All of lore.kernel.org
 help / color / mirror / Atom feed
From: David Laight <David.Laight@ACULAB.COM>
To: 'Andrew Morton' <akpm@linux-foundation.org>,
	Jani Nikula <jani.nikula@intel.com>
Cc: "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"intel-gfx@lists.freedesktop.org"
	<intel-gfx@lists.freedesktop.org>,
	"dri-devel@lists.freedesktop.org"
	<dri-devel@lists.freedesktop.org>,
	"Christian König" <christian.koenig@amd.com>,
	"David Gow" <davidgow@google.com>
Subject: RE: [PATCH 0/4] log2: make is_power_of_2() more generic
Date: Thu, 30 Mar 2023 21:53:03 +0000	[thread overview]
Message-ID: <549987e4967d45159573901d330c96a0@AcuMS.aculab.com> (raw)
In-Reply-To: <20230330125041.83b0f39fa3a4ec1a42dfd95f@linux-foundation.org>

From: Andrew Morton
> Sent: 30 March 2023 20:51
> 
> On Thu, 30 Mar 2023 13:42:39 +0300 Jani Nikula <jani.nikula@intel.com> wrote:
> 
> > is_power_of_2() only works for types <= sizeof(unsigned long) and it's
> > also not a constant expression. There are a number of places in kernel
> > where is_power_of_2() is called on u64, which fails on 32-bit
> > builds. Try to remedy that. While at it, make it a constant expression
> > when possible.
> 
> Yes, the current `is_power_of_2(unsigned long n)' isn't very general.
> 
> But wouldn't all these issues be addressed by simply doing
> 
> #define is_power_of_2(n) (n != 0 && ((n & (n - 1)) == 0))
> 
> ?
> 
> (With suitable tweaks to avoid evaluating `n' more than once)

I think you need to use the 'horrid tricks' from min() to get
a constant expression from constant inputs.

For non-constants this looks ok (see https://godbolt.org/z/G73MTr9jn)

	David

static inline int lisp2(unsigned long n)
{
    return n && !(n & (n - 1));
}

static inline int llisp2(unsigned long long lln)
{
#if 0  // I think this looks worse, esp. for gcc on x86
    return lln && !(lln & (lln - 1));
#else
    unsigned long n = lln;
    if (lln >= 1ull << 32) {
        if (n)
            return 0;
        n = lln >> 32; 
    }
    return lisp2(n);
#endif
}

#define isp2(n) (sizeof ((n)+0) == sizeof (long) ? lisp2(n) : llisp2(n))

int is12(unsigned long long  i)
{
    return isp2(i);
}

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)


WARNING: multiple messages have this Message-ID (diff)
From: David Laight <David.Laight@ACULAB.COM>
To: 'Andrew Morton' <akpm@linux-foundation.org>,
	Jani Nikula <jani.nikula@intel.com>
Cc: "David Gow" <davidgow@google.com>,
	"intel-gfx@lists.freedesktop.org"
	<intel-gfx@lists.freedesktop.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"dri-devel@lists.freedesktop.org"
	<dri-devel@lists.freedesktop.org>,
	"Christian König" <christian.koenig@amd.com>
Subject: RE: [PATCH 0/4] log2: make is_power_of_2() more generic
Date: Thu, 30 Mar 2023 21:53:03 +0000	[thread overview]
Message-ID: <549987e4967d45159573901d330c96a0@AcuMS.aculab.com> (raw)
In-Reply-To: <20230330125041.83b0f39fa3a4ec1a42dfd95f@linux-foundation.org>

From: Andrew Morton
> Sent: 30 March 2023 20:51
> 
> On Thu, 30 Mar 2023 13:42:39 +0300 Jani Nikula <jani.nikula@intel.com> wrote:
> 
> > is_power_of_2() only works for types <= sizeof(unsigned long) and it's
> > also not a constant expression. There are a number of places in kernel
> > where is_power_of_2() is called on u64, which fails on 32-bit
> > builds. Try to remedy that. While at it, make it a constant expression
> > when possible.
> 
> Yes, the current `is_power_of_2(unsigned long n)' isn't very general.
> 
> But wouldn't all these issues be addressed by simply doing
> 
> #define is_power_of_2(n) (n != 0 && ((n & (n - 1)) == 0))
> 
> ?
> 
> (With suitable tweaks to avoid evaluating `n' more than once)

I think you need to use the 'horrid tricks' from min() to get
a constant expression from constant inputs.

For non-constants this looks ok (see https://godbolt.org/z/G73MTr9jn)

	David

static inline int lisp2(unsigned long n)
{
    return n && !(n & (n - 1));
}

static inline int llisp2(unsigned long long lln)
{
#if 0  // I think this looks worse, esp. for gcc on x86
    return lln && !(lln & (lln - 1));
#else
    unsigned long n = lln;
    if (lln >= 1ull << 32) {
        if (n)
            return 0;
        n = lln >> 32; 
    }
    return lisp2(n);
#endif
}

#define isp2(n) (sizeof ((n)+0) == sizeof (long) ? lisp2(n) : llisp2(n))

int is12(unsigned long long  i)
{
    return isp2(i);
}

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)


WARNING: multiple messages have this Message-ID (diff)
From: David Laight <David.Laight@ACULAB.COM>
To: 'Andrew Morton' <akpm@linux-foundation.org>,
	Jani Nikula <jani.nikula@intel.com>
Cc: "David Gow" <davidgow@google.com>,
	"intel-gfx@lists.freedesktop.org"
	<intel-gfx@lists.freedesktop.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"dri-devel@lists.freedesktop.org"
	<dri-devel@lists.freedesktop.org>,
	"Christian König" <christian.koenig@amd.com>
Subject: Re: [Intel-gfx] [PATCH 0/4] log2: make is_power_of_2() more generic
Date: Thu, 30 Mar 2023 21:53:03 +0000	[thread overview]
Message-ID: <549987e4967d45159573901d330c96a0@AcuMS.aculab.com> (raw)
In-Reply-To: <20230330125041.83b0f39fa3a4ec1a42dfd95f@linux-foundation.org>

From: Andrew Morton
> Sent: 30 March 2023 20:51
> 
> On Thu, 30 Mar 2023 13:42:39 +0300 Jani Nikula <jani.nikula@intel.com> wrote:
> 
> > is_power_of_2() only works for types <= sizeof(unsigned long) and it's
> > also not a constant expression. There are a number of places in kernel
> > where is_power_of_2() is called on u64, which fails on 32-bit
> > builds. Try to remedy that. While at it, make it a constant expression
> > when possible.
> 
> Yes, the current `is_power_of_2(unsigned long n)' isn't very general.
> 
> But wouldn't all these issues be addressed by simply doing
> 
> #define is_power_of_2(n) (n != 0 && ((n & (n - 1)) == 0))
> 
> ?
> 
> (With suitable tweaks to avoid evaluating `n' more than once)

I think you need to use the 'horrid tricks' from min() to get
a constant expression from constant inputs.

For non-constants this looks ok (see https://godbolt.org/z/G73MTr9jn)

	David

static inline int lisp2(unsigned long n)
{
    return n && !(n & (n - 1));
}

static inline int llisp2(unsigned long long lln)
{
#if 0  // I think this looks worse, esp. for gcc on x86
    return lln && !(lln & (lln - 1));
#else
    unsigned long n = lln;
    if (lln >= 1ull << 32) {
        if (n)
            return 0;
        n = lln >> 32; 
    }
    return lisp2(n);
#endif
}

#define isp2(n) (sizeof ((n)+0) == sizeof (long) ? lisp2(n) : llisp2(n))

int is12(unsigned long long  i)
{
    return isp2(i);
}

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)


  reply	other threads:[~2023-03-30 21:53 UTC|newest]

Thread overview: 41+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-03-30 10:42 [PATCH 0/4] log2: make is_power_of_2() more generic Jani Nikula
2023-03-30 10:42 ` [Intel-gfx] " Jani Nikula
2023-03-30 10:42 ` Jani Nikula
2023-03-30 10:42 ` [PATCH 1/4] log2: add helper __IS_POWER_OF_2() Jani Nikula
2023-03-30 10:42   ` [Intel-gfx] " Jani Nikula
2023-03-30 10:42   ` Jani Nikula
2023-03-30 10:42 ` [PATCH 2/4] log2: have is_power_of_2() support bigger types than unsigned long Jani Nikula
2023-03-30 10:42   ` [Intel-gfx] " Jani Nikula
2023-03-30 10:42   ` Jani Nikula
2023-03-30 10:42 ` [PATCH 3/4] log2: allow use of is_power_of_2() in constant expressions Jani Nikula
2023-03-30 10:42   ` [Intel-gfx] " Jani Nikula
2023-03-30 10:42   ` Jani Nikula
2023-03-30 10:42 ` [PATCH 4/4] drm/i915/reg: use is_power_of_2() from log2.h Jani Nikula
2023-03-30 10:42   ` [Intel-gfx] " Jani Nikula
2023-03-30 10:42   ` Jani Nikula
2023-03-30 10:59 ` [PATCH 0/4] log2: make is_power_of_2() more generic Christian König
2023-03-30 10:59   ` [Intel-gfx] " Christian König
2023-03-30 10:59   ` Christian König
2023-03-30 11:10 ` [Intel-gfx] ✗ Fi.CI.CHECKPATCH: warning for " Patchwork
2023-03-30 11:10 ` [Intel-gfx] ✗ Fi.CI.SPARSE: " Patchwork
2023-03-30 11:23 ` [Intel-gfx] ✓ Fi.CI.BAT: success " Patchwork
2023-03-30 19:50 ` [PATCH 0/4] " Andrew Morton
2023-03-30 19:50   ` [Intel-gfx] " Andrew Morton
2023-03-30 19:50   ` Andrew Morton
2023-03-30 21:53   ` David Laight [this message]
2023-03-30 21:53     ` [Intel-gfx] " David Laight
2023-03-30 21:53     ` David Laight
2023-03-30 22:18     ` Andrew Morton
2023-03-30 22:18       ` Andrew Morton
2023-03-30 22:18       ` [Intel-gfx] " Andrew Morton
2023-03-31  7:33       ` David Laight
2023-03-31  7:33         ` [Intel-gfx] " David Laight
2023-03-31  7:33         ` David Laight
2023-03-31  8:31       ` Jani Nikula
2023-03-31  8:31         ` [Intel-gfx] " Jani Nikula
2023-03-31  8:31         ` Jani Nikula
2023-04-05 15:27         ` Steven Price
2023-04-05 15:27           ` [Intel-gfx] " Steven Price
2023-04-05 15:27           ` Steven Price
2024-04-12 10:01           ` Jani Nikula
2023-03-31  5:58 ` [Intel-gfx] ✓ Fi.CI.IGT: success for " Patchwork

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=549987e4967d45159573901d330c96a0@AcuMS.aculab.com \
    --to=david.laight@aculab.com \
    --cc=akpm@linux-foundation.org \
    --cc=christian.koenig@amd.com \
    --cc=davidgow@google.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=jani.nikula@intel.com \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.